Field notes

The reading room.

Everything Contextkeeping publishes, in one place: free, no email required. The essays carry the doctrine; the instruments make it runnable; the standard makes it checkable.

The essays

• KCS was built for human writers and human readers. Both are gone.the founding essay

  Why both of the methodology's founding assumptions expired, and what a governed knowledge operation looks like now.

• Answer Economicsthe measurement doctrine

  Deflection measures what didn't happen. Cost per answer measures what did, across six delivery channels.

• The Operating Modelthe executive view

  Roles, gates, cadence, and the one number the program answers to, written for the person who owns the budget.

• The language of knowledge work is failing its own testthe vocabulary layer

  Terms are retrieval keys. Terminology governance as the vocabulary layer of machine-ready knowledge.

• The Content Graduation Modelthe movement layer

  Content earns trust rung by rung, retires on evidence, improves the product that generated it, and becomes the knowledge layer that grounds AI.

The standard

• Machine-Ready Knowledge (MRK-1.0)free to adopt

  Four clauses for articles both readers can trust, with a served JSON Schema you can validate against today.

• MRK-2.0, the maturity layerreleased 2026-10-19

  The floor is not the ceiling. The four clauses get an article to L1; MRK-2.0 climbs above that line without leaving Markdown. The Structure Graduation Ladder scores a corpus L0–L4 and names each article's next edit, Clause §R governs the reader that assembles answers, and the graph projection turns the corpus into queryable RDF with one served JSON-LD context.

The build log

The essays argue; the reference implementation demonstrates. The build log walks through the system I actually run: real files, real commits, real screenshots.

• 01 · The git-backed system of recordverified 2026-09-08

  The record is a git repository: every capture, accept, validate, and archive is a commit made at the gate, and the AI layer is a disposable projection of it.

• 02 · The validation gate in actionverified 2026-09-08

  The schema gate that refuses malformed knowledge, the lifecycle that never resurrects the dead, and the R1–R6 rejection codes that turn every "no" into telemetry.

• 03 · The gap loop, on a clockverified 2026-09-08

  A refused question becomes a logged gap, a capture request, and a validated object in under a day, and the object cites the gap as its source.

• 04 · Two kinds of oldverified 2026-09-08

  Old is not stale: volatility routes human attention, dates move only on real re-verification, and genuine staleness gets a lifecycle act, never a bump.

The instruments, free

• One issue, end to endthe loop, demonstrated

  A single support ticket riding the whole Governed Knowledge Loop, about seven minutes of human judgment in its entire life.

• The maturity modelone-page self-assessment

  Place your program honestly; your level is your weakest dimension.

• The terminology one-pagertwelve swaps, one page

  The twelve highest-yield term swaps and the four-move program.

• The graduation one-pagerthe model, one page

  The trust ladder, the four signal families, and the 1-flow / 5-candidate / 2-pass pilot.

The estate log

What shipped, when: history, not freshness claims. Entries are added when something real lands; no cadence is promised, and verified: dates elsewhere on the estate move only when a human actually re-verifies the content.

• 2026-10-19 MRK-2.0 released

  The maturity layer over the MRK-1.1 article format: five producer classes (L0–L4), the graph projection and SHACL gate, declared reuse with a drift gate, Clause §R for consumers, and sealed conformance evidence, with one offline kit that runs all of it. Articles do not change: MRK-1.0 and MRK-1.1 are frozen, and there is no new mrk_version.

• 2026-09-18 The maturity layer (MRK-1.1 → 2.0)

  The additive layer above the four clauses: the Structure Graduation Ladder (L0–L4), Clause §R for retrieval, and the graph projection, with the 1.1 schema and JSON-LD context served free.

• 2026-09-08 One issue, end to end

  The free walkthrough: the Governed Knowledge Loop on a single ticket.

• 2026-09-08 The MRK specimen

  One real article conforming to MRK-1.0, published with the command that proves it.

• 2026-09-08 Build logs 01–04

  The reference implementation, shown running: the record, the gate, the gap loop, and the two kinds of old.

• 2026-09-07 The AnswerEcon specimen restored

  The audit deliverable back on its own domain, readable in full: queue closed, waitlist open.

• 2026-09-07 The Content Graduation Model

  Content earns trust rung by rung, retires on evidence, and grounds AI.

• 2026-09-05 The doctrine, re-verified and expanded

  "KCS doesn't need a successor. It needs an expansion." A real re-verification, which is the only kind that moves a date here.

An occasional letter

When new doctrine lands (a new essay, a new instrument, a revision worth knowing about), I write a short letter. Occasional means occasional: no cadence promises, no digest padding, one-click unsubscribe.

The reading requires no email; the letter is optional by design. Questions or shop talk: hello@contextkeeping.com.